Recognition of Wood Defects Based on Artificial Neural Network

The recognition model of wood defects based on artificial neural network (ANN) was presented for classifying the defects effectively. X-ray was adopted as a measure method for log nondestructive testing. Applying MATLAB and VC++ image processing program processed the image of log with defects and extracted the characters of the image. The mathematic model of defects recognition was established according to characteristic parameters. So back propagating networks was constructed. In this paper, three common defects which are knot, grub-hole and rot were studied. The experimental results show that ANN is an effective method for the nondestructive testing and classifying of three defects. This method also can be used in other log defects nondestructive testing and classifying.